Evaluating the effectiveness of machine translation of audio description: the results of two pilot studies in the English-Dutch language pair

The field of translation is undergoing various profound changes. On the one hand it being thoroughly reshaped by advent and constant improvement new technologies. other hand, forms are starting to see light day in wake social legal developments that require products content created, accessible for everybody. One these translation, audio description (AD), a service aimed at making audiovisual people with sight loss. New legislation requires this 2025, which constitutes tremendous task given limited number present trained as describers. A possible solution would be use machine translate existing descriptions into different languages. Since AD characterized short sentences simple, concrete language, could good candidate translation. In study, we want test hypothesis English-Dutch language pair. Three 30 minute excerpts Dutch movies were originally described English, translated using DeepL. translations analysed harmonized DQF-MQM error typology taking account specific multimodal nature source text intersemiotic dimension original process. analysis showed MT output had relatively high rate, particularly categories Accuracy – mistranslation Fluency grammar. This seems indicate extensive post-editing will needed, before can used professional context.